Name: HORUS
Availability: InStock

IMU Reader (Python)

Reads a 6-axis IMU (accelerometer + gyroscope) at 100 Hz and publishes Imu messages for downstream consumers. Includes simple gyro integration for orientation estimation.

Problem

You need to read IMU hardware at a fixed rate, validate readings, and publish typed Imu messages from Python.

When To Use

Reading a 6-axis or 9-axis IMU over I2C/SPI from Python
Publishing raw Imu messages for fusion, logging, or SLAM nodes
Prototyping IMU pipelines before porting to Rust

Prerequisites

HORUS installed (Installation Guide)
Python 3.8+ with horus package

horus.toml

[package]
name = "imu-reader-py"
version = "0.1.0"
description = "100 Hz IMU sensor with orientation publishing"
language = "python"

Complete Code

#!/usr/bin/env python3
"""100 Hz IMU reader with gyro-integrated orientation estimation."""

import math
import horus
from horus import Node, Scheduler, Imu, us, ms

# ── State ────────────────────────────────────────────────────

roll = [0.0]
pitch = [0.0]
yaw = [0.0]
tick_count = [0]

# ── Hardware stub ────────────────────────────────────────────

def read_hardware():
    """Read IMU hardware — replace with your I2C/SPI driver.

    Returns:
        horus.Imu with accel and gyro fields populated.
    """
    return Imu(
        accel_x=0.0,
        accel_y=0.0,
        accel_z=9.81,       # gravity on Z axis
        gyro_x=0.01,        # slow roll
        gyro_y=0.0,
        gyro_z=0.05,        # slow yaw rotation
    )

# ── Node callbacks ───────────────────────────────────────────

def imu_tick(node):
    imu = read_hardware()

    # Validate before publishing — hardware faults produce NaN
    if (math.isnan(imu.accel_x) or math.isnan(imu.accel_y) or
        math.isnan(imu.accel_z) or math.isnan(imu.gyro_x) or
        math.isnan(imu.gyro_y) or math.isnan(imu.gyro_z)):
        return  # skip corrupted readings

    # Publish raw IMU for any subscriber
    node.send("imu.raw", imu)

    # Simple gyro integration (replace with Madgwick/Mahony in production)
    dt = 1.0 / 100.0  # 100 Hz -> 10 ms per tick
    roll[0] += imu.gyro_x * dt
    pitch[0] += imu.gyro_y * dt
    yaw[0] += imu.gyro_z * dt

    tick_count[0] += 1

    # Publish orientation estimate as a dict (or define a custom class)
    node.send("imu.orientation", {
        "roll": roll[0],
        "pitch": pitch[0],
        "yaw": yaw[0],
        "tick": tick_count[0],
    })

def imu_shutdown(node):
    print(f"ImuReader: {tick_count[0]} ticks, final yaw={yaw[0]:.4f} rad")

# ── Main ─────────────────────────────────────────────────────

imu_node = Node(
    name="ImuReader",
    tick=imu_tick,
    shutdown=imu_shutdown,
    rate=100,                      # 100 Hz sensor rate
    order=0,
    pubs=["imu.raw", "imu.orientation"],
    subs=[],
    budget=800 * us,               # 800 us budget for I2C/SPI read
)

if __name__ == "__main__":
    horus.run(imu_node)

Expected Output

[HORUS] Scheduler running — tick_rate: 1000 Hz
[HORUS] Node "ImuReader" started (100 Hz)
^C
ImuReader: 500 ticks, final yaw=2.5000 rad
[HORUS] Shutting down...
[HORUS] Node "ImuReader" shutdown complete

Key Points

horus.Imu is the typed message with accel_x/y/z and gyro_x/y/z fields
NaN validation catches hardware faults before they propagate to downstream nodes
Gyro integration drifts over time — in production, use a complementary or Madgwick filter
read_hardware() is a placeholder — replace with your actual I2C/SPI driver (e.g., smbus2 for Linux)
budget=800 * us accounts for I2C bus latency on typical embedded hardware

Variations

Complementary filter: Blend accelerometer gravity direction with gyro integration for stable pitch/roll
Hardware I2C: Use smbus2 to read from an MPU6050 or BNO055 over I2C
High-rate IMU: Set rate=1000 with budget=200 * us for 1 kHz industrial IMUs

Common Errors

Symptom	Cause	Fix
Orientation drifts continuously	Pure gyro integration without correction	Use complementary or Madgwick filter
`NaN` in published data	Hardware fault or I2C read error	Add NaN check before publishing
Orientation jumps on startup	Initial gyro bias not calibrated	Average first 100 readings as bias offset
Yaw rotates when stationary	Gyro bias in Z-axis	Subtract calibrated bias from `gyro_z`
Topic data stale	Publisher rate too low	Verify `rate=100` matches hardware capability